Genetic studies have revealed numerous non-coding sequence variants associated with diabetes or obesity risk. Generating reference 3D epigenome data in in key metabolic cell or tissue types, such as human pancreatic ? cell, is therefore very important for the understanding of disease etiology. Conventional genomic methods to study transcriptional regulation include RNA-seq and ChIP-seq, for the mapping of transcriptome and epigenome, respectively. These technologies typically require ~1 million cells per assay. Hi-C is currently the most popular method to map the 3D genome organization in an unbiased fashion, but tens of millions cells are required to achieve kilobase resolution 3D genome analysis. On the other hand, the ? cell research community is relying on islet samples from deceased donors, which are precious and very expensive. Low-input technologies are therefore essential for ? cell genomic studies. The overall goal of this project is to combine several state-of-art single-cell and low- input genomic technologies to systematically characterize the ? cell 3D enhancer regulome from a cohort of 40 fresh human islet samples; some key technologies are the original inventions from the laboratories of our team. In aim 1, we will use a massively parallel single cell RNA-seq method (Drop-seq) to generate a comprehensive dataset of single islet cell transcriptome, and simultaneously use a low-input ChIPmentation method to map enhancers and promoters from a cohort of 40 human islet donors. This will lead to the identification of diabetes and obesity signature genes and variable enhancer loci (VELs) correlated with disease status. In aim 2, we will map the human ? cell 3D genome at kilobase resolution using a highly efficient easy Hi-C (eHi-C) method. The map will reveal all the interactions between individual enhancers and promoters. In aim 3, we will for the first time perform an in-depth study of the 3D regulome at the human INS locus, and quantify the activity of all enhancers near INS using a haplotype-resolved human cell line. We will also test a hypothesis that MAU2-NIPBL cohesin loading complex may regulate insulin gene expressions through mediating enhancer-promoter DNA looping at this locus. Finally, we will use a novel high-throughput Mosaic-seq method to validation the in vivo activity of dozens of ? cell enhancers at single cell level. This comprehensive 3D regulome data will provide a key resource for the understanding of the functions of non-coding genome in T2D or obesity.